Pasquale, Eduardo and Ming would like to hold a discussion session
on the recent Planck result, including paper1,paper2,paper3,paper4
and paper5, especially on the X-ray/SZ comparison and the
optical mass proxy.
The topics for discussion may include (but not limited to):

  • What can Planck contribute most?

  • ICM seen by SZ vs. by X-rays

  • Mass proxies in optical, SZ and X-rays and the cross-calibration

We would like to start it at 2PM of Wednesday in Founders' room, after the
lunch talk by Pasquale.

Summary of Discussion

The first part of the discussion was mostly a summary of the work that I (Eduardo) have made concerning the Planck maxBCG
results. We considered many possibilities for the source of the discrepancy.

One possibility is whether properly propagaing all selection effects into the scaling relations could alleviate the tension. I (Eduardo)
talked about work I've done using Monte Carlo realizations that demonstrate this is not the case, so it's not selection effects, assuming
log-normal scatter is an adequate description.

Steve pointed out this still leaves the possibility of projection effects. My opinion (Eduardo) is that this is unlikely, since we estimate the
fraction of clusters to be projection to be something around 5%-10% at the high mass end. Moreover, if projections were the source of the
problem for the stacked SZ signal, I would have expected the Lx measurements to also suffer, so that the Lx-M relation should be off too.
This is far from a fool-proof argument though, so projection effects might play a role, but one needs to explain then why Lx-M comes out
right.

Another possible explanation for the tension is that the maxBCG mass calibration is wrong. While possible, this creates other problems,
namely the comsology from maxBCG would drop to sigma_8=0.71, and the Lx-M relation from the stack measurements, which now
agrees for optical and SZ, would change, i.e. fixing the SZ scaling would ruin the Lx scaling. So, possible, but creates new problems
that would then have to be fixed.


The possibility that got the most attention is correlated scatter. Here, one needs to be careful. If we write the observed signal as

Yobs = Ytrue + Ynoise

then the correlation can be between richness and Ytrue, or between richness and Ynoise. I have considered the former possibility,
but find that this really doesn't help at all, in large part because the scatter from noise is much, much larger than the intrinsic scatter
in the Ysz-M relation. The latter *might* expalin things, and this was the focus of a lot of the discussion with many people contributing.
The reason that Ynoise can be correlated with richness is that Planck sets the size of a cluster based on the richness data, and the Planck
data has a strong size-signal degeneracy. It is possible that this souce of correlated scatter could result in a bias. An interesting suggestion
by Andrey was to make the following plot: take the maxBCG clusters thata are also in the MCXC catalog, and assign them a size based on
the X-ray data, and re-estimate Ysz. One could make a plot of the SZ signal estimated using the X-ray size vs the SZ signal estiamted
using the optical size. Differences in this plot could signal that this might be the source of the discrepancy, but it's still unclear.

One last thing tha was touched was that from maxBCG data,one should expect the X-ray bright maxBCG clusters to have a higher
SZ signal than the full sample, so the fact that the X-ray bright objects have a higher SZ signal is expected. In that sense,
the apparent agreement in Fig. 5 of the Planck paper could well be a coincidence.